Topic : 
無人機於未知環境之影像輔助降落系統
Visual Aid System for UAV Landing in Unsurveyed Environments

Speaker :  
鄭琇文
國立交通大學 機械工程學系博士候選人

Date & Location : 2020/06/04 (Thu) 3:30 pm - 4:30 pm
Engineering 5 Building B1 International Conference Hall (工程五館 B1 國際會議廳)


Abstract:  
UAVs are currently used for various missions, ranging from
military to civil and commercial. In order to reduce the landing
risk, we attempt to develop a vision system as an aid for UAV
landing in unsurveyed environments. For the rotorcraft UAVs,
we introduced the self-supervised learning method to assess
the landing risk by evaluating the planarity of terrain
appearance. Once the safety of the landing spot was confirmed,
the visual landing process was engaged through a hybrid optical
flow framework. As for the fixed-wing UAV category, we
proposed an adaptive runway detection algorithm to improve
the recognition success rate. In combination with the onboard
inertial measurements, the lateral deviation used for landing
path correction can be further estimated through the vision-
based results.